Towards a Complete Pipeline for Segmenting Nuclei in Feulgen-Stained Images
Luiz Antonio Buschetto Macarini, Aldo von Wangenheim, Felipe Perozzo Daltoé, Alexandre Sherlley Casimiro Onofre, Fabiana Botelho de Miranda Onofre, Marcelo Ricardo Stemmer · Anais do XI Computer on the Beach - COTB '20 · 2020
Cervical cancer is the second most common cancer type in womenaround the world. In some countries, due to non-existent or inadequatescreening, it is often detected at late stages, making standardtreatment options often absent or unaffordable. It is a deadlydisease that could benefit from early detection approaches. It isusually done by cytological exams which consist of visually inspectingthe nuclei searching for morphological alteration. Since itis done by humans, naturally, some subjectivity is introduced. Computationalmethods could be used to reduce this, where the firststage of the process would be the nuclei segmentation. In this context,we present a complete pipeline for the segmentation of nucleiin Feulgen-stained images using Convolutional Neural Networks.Here we show the entire process of segmentation, since the collectionof the samples, passing through pre-processing, training thenetwork, post-processing and results evaluation. We achieved anoverall IoU of 0.78, showing the affordability of the approach of nucleisegmentation on Feulgen-stained images. The code is availablein: https://github.com/luizbuschetto/feulgen_nuclei_segmentation